290 views
0 0 votes

PCA is run on a dataset that has 2 features. The resulting principal components are $\mathbf{w}_1$ and $\mathbf{w}_2$. We represent the points in 2D space in terms of this new coordinate system made up of the principal components. The first coordinate corresponds to $\mathbf{w}_1$ and the second to $\mathbf{w}_2$. In such a scenario, what would be the sign of the coordinates for the points $P$ and $Q$ ?
 

 

  1. P: (+ve,-ve)
     
  2. P: (-ve,+ve)
     
  3. Q: (-ve,+ve)
     
  4. Q: (+ve,-ve)

2 Answers

0 0 votes

1. Each principal component $w$ is a direction.

Its coordinate for a point is the dot product of that point with $w$.

The sign of this dot product tells us on which "side" of the origin the point lies when measured along $w$.

2. To visualize the sign, draw the line perpendicular to $w$ through the origin.

This line divides the plane into two half-planes:

  • The half-plane that contains $w$ itself gives positive values of $w^T x$.
     
  • The opposite half-plane gives negative values.

3. Apply this to each vector.

  • For $w_1$ : check whether $P$ and $Q$ lie on the same side as the arrow of $w_1$ (positive) or the opposite side (negative).
     
  • For $\mathrm{w}_2$ : do the same with the perpendicular line to $\mathrm{w}_2$.

Following this rule:

  • P lies opposite $w_1 \rightarrow w_1$ coordinate negative; same side as $w_2 \rightarrow w_2$ coordinate positive.
     
  • Q lies same side as $\mathrm{w}_1 \rightarrow \mathrm{w}_1$ coordinate positive; opposite $\mathrm{w}_2 \rightarrow \mathrm{w}_2$ coordinate negative.

That's why the signs come out as:

  • $\mathrm{P}=(-,+)$
     
  • $\mathrm{Q}=(+,-)$
Answer:
Position:
Show:

Related questions

0 0 votes
1 1 answer
299
299 views
GO Classes asked Sep 23, 2025
299 views
Let $x_1, x_2, \ldots, x_n$ be $d$-dimensional data points $(d>n)$ and $X$ be the matrix of shape $d \times n$ containing the data points. The $k^{\text {th }}$ largest e...
1 1 vote
2 2 answers
384
384 views
GO Classes asked Sep 23, 2025
384 views
Consider a dataset that has $1000$ samples, where each sample belongs to $\mathbb{R}^{30}$. PCA is run on this dataset and the top $4$ principal components are retained, ...
0 0 votes
2 2 answers
339
339 views
GO Classes asked Sep 23, 2025
339 views
Consider the following images of points in 2D space. The red line segments in one of the images represent the lengths of the residues after projecting the points on the l...
3 3 votes
2 2 answers
378
378 views
GO Classes asked Sep 23, 2025
378 views
Which of the following best represents a valid voronoi diagram for K -means algorithm with $\mathrm{K}=3$ ? (The dots represent the cluster centres of respective clusters...